Part1: datasets “People Receiving Homeless Response Services”

Description of datasets

The demo datasets are defined as “People Receiving Homeless Response Services by Age, Race, Ethnicity, and Gender”.

This data set gives annual statistics by age group, race, ethnicity, and gender on the number of people who have benefited from homelessness assistance services.

The Homelessness Data Integration System (HDIS), an extensive data repository that compiles and examines information from each of California’s 44 Continuums of Care (CoC), is the source of this information. Every Community of Communities (CoC) collects and disseminates information on the people it serves via a range of programs, such as those aimed at ending homelessness, offering outreach services to the homeless, assisting with permanent housing solutions, and other projects in line with California’s Housing First objectives.

  • Important Variables:

CALENDAR_YEAR: Calendar Year
LOCATION: Location
ETHNICITY: Ethnicity
AGE_GROUP_PUBLIC: Age Group
GENDER: Gender
RACE: Race
EXPERIENCING_HOMELESSNESS: Number of people receiving homelessness services

Data Process

First, we converted all the variable names to lowercase.

Then we removed the unneeded variable “location_id”, and removed the null value of “experiencing_homelessness”.

Finally, we batch export and import the four cleaned csv files into datasets.

# import data
library(tidyverse)
age_demo = read.csv("./original data/demo/age_demo.csv")
ethnicity_demo = read.csv("./original data/demo/ethnicity_demo.csv")
gender_demo = read.csv("./original data/demo/gender_demo.csv")
race_demo = read.csv("./original data/demo/race_demo.csv")
head(age_demo) 
##   CALENDAR_YEAR LOCATION_ID   LOCATION AGE_GROUP_PUBLIC
## 1          2017         All California            18-24
## 2          2017         All California            25-34
## 3          2017         All California            35-44
## 4          2017         All California            45-54
## 5          2017         All California            55-64
## 6          2017         All California              65+
##   EXPERIENCING_HOMELESSNESS
## 1                     15984
## 2                     28654
## 3                     25831
## 4                     27651
## 5                     23396
## 6                      7111
head(ethnicity_demo)
##   CALENDAR_YEAR LOCATION_ID               LOCATION           ETHNICITY
## 1          2017         All             California     Hispanic/Latinx
## 2          2017         All             California Not Hispanic/Latinx
## 3          2017         All             California             Unknown
## 4          2017      CA-500 Santa Clara County CoC     Hispanic/Latinx
## 5          2017      CA-500 Santa Clara County CoC Not Hispanic/Latinx
## 6          2017      CA-500 Santa Clara County CoC             Unknown
##   EXPERIENCING_HOMELESSNESS
## 1                     60962
## 2                    119153
## 3                      3791
## 4                      5034
## 5                      5074
## 6                        57
head(gender_demo)
##   CALENDAR_YEAR LOCATION_ID   LOCATION              GENDER
## 1          2017         All California              Female
## 2          2017         All California                Male
## 3          2017         All California Non-Singular Gender
## 4          2017         All California  Questioning Gender
## 5          2017         All California         Transgender
## 6          2017         All California             Unknown
##   EXPERIENCING_HOMELESSNESS
## 1                     79670
## 2                    101901
## 3                       148
## 4                         *
## 5                       676
## 6                      1505
head(race_demo)
##   CALENDAR_YEAR LOCATION_ID   LOCATION
## 1          2017         All California
## 2          2017         All California
## 3          2017         All California
## 4          2017         All California
## 5          2017         All California
## 6          2017         All California
##                                            RACE EXPERIENCING_HOMELESSNESS
## 1 American Indian, Alaska Native, or Indigenous                      5638
## 2                       Asian or Asian American                      3005
## 3           Black, African American, or African                     57665
## 4                                Multiple Races                      9048
## 5           Native Hawaiian or Pacific Islander                      2555
## 6                                       Unknown                      6391
# Clean data
library(readr)
folder_path <- "./original data/demo/"
csv_files <- list.files(folder_path, pattern = ".csv")

# define a function to process a CSV file
clean_csv <- function(input_file, output_file) {
  data <- read.csv(input_file)
  data_cleaned <- data |> 
    filter(EXPERIENCING_HOMELESSNESS != "*") |>
    select(-LOCATION_ID) |>
    janitor::clean_names() 
   write.csv(data_cleaned, file = output_file, row.names = FALSE)
   #return(data_cleaned)
}

output_folder <- "./processed data/demo/"

# process CSV files using the lapply function
cleaned_data_list <- lapply(csv_files, function(file) {
  input_file <- paste0(folder_path, file)
  output_file <- paste0(output_folder, "cleaned_", file)
  clean_csv(input_file, output_file)
})

# import clean data
age_clean = read.csv("./processed data/demo/cleaned_age_demo.csv")
ethnicity_clean = read.csv("./processed data/demo/cleaned_ethnicity_demo.csv")
gender_clean = read.csv("./processed data/demo/cleaned_gender_demo.csv")
race_clean = read.csv("./processed data/demo/cleaned_race_demo.csv")

Summary

To begin with the analyzing process, we first summarize the counts and percentages of Californian homeless people from 2017 to 2023, stratified by age, ethnicity, gender, race, separately for each of the four datasets:

Dataset categorized by age:

age_clean_for_summary = 
  age_clean |>
  select(-location)

age_clean_for_summary |>
  group_by(age_group_public) |>
  summarize(Homelessness = sum(experiencing_homelessness)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
age_group_public Homelessness Percent
18-24 281727 7.886308
25-34 576530 16.138649
35-44 566387 15.854719
45-54 524878 14.692769
55-64 493348 13.810158
65+ 189685 5.309801
Under 18 813273 22.765732
Unknown 126528 3.541864

Dataset categorized by ethnicity:

ethnicity_clean_for_summary = 
  ethnicity_clean |>
  select(-location)

ethnicity_clean_for_summary |>
  group_by(calendar_year) |>
  summarize(Homelessness = sum(experiencing_homelessness)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
calendar_year Homelessness Percent
2017 373281 10.63152
2018 434563 12.37691
2019 486049 13.84330
2020 514756 14.66091
2021 547527 15.59427
2022 640070 18.23001
2023 514832 14.66307 
ethnicity_clean_for_summary |>
  group_by(ethnicity) |>
  summarize(Homelessness = sum(experiencing_homelessness)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
ethnicity Homelessness Percent
Hispanic/Latinx 1224078 34.863310
Not Hispanic/Latinx 2187297 62.297021
Unknown 99703 2.839669

Dataset categorized by gender:

gender_clean_for_summary = 
  gender_clean |>
  select(-location)

gender_clean_for_summary |>
  group_by(calendar_year) |>
  summarize(Homelessness = sum(experiencing_homelessness)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
calendar_year Homelessness Percent
2017 373071 10.63095
2018 434374 12.37783
2019 485807 13.84346
2020 514506 14.66126
2021 547260 15.59461
2022 639749 18.23015
2023 514523 14.66174 
gender_clean_for_summary |>
  group_by(gender) |>
  summarize(Homelessness = sum(experiencing_homelessness)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
gender Homelessness Percent
Female 1479598 42.1623177
Male 1969207 56.1141143
Non-Singular Gender 6170 0.1758190
Questioning Gender 161 0.0045878
Transgender 15788 0.4498916
Unknown 38366 1.0932696

Dataset categorized by race:

race_clean_for_summary = 
  race_clean |>
  select(-location)

race_clean_for_summary |>
  group_by(calendar_year) |>
  summarize(Homelessness = sum(experiencing_homelessness)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
calendar_year Homelessness Percent
2017 373049 10.62857
2018 434372 12.37573
2019 485876 13.84314
2020 514587 14.66115
2021 547394 15.59585
2022 639935 18.23245
2023 514656 14.66311 
race_clean_for_summary |>
  group_by(race) |>
  summarize(Homelessness = sum(experiencing_homelessness)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
race Homelessness Percent
American Indian, Alaska Native, or Indigenous 108770 3.098976
Asian or Asian American 62782 1.788728
Black, African American, or African 1026875 29.256790
Multiple Races 153850 4.383355
Native Hawaiian or Pacific Islander 44051 1.255061
Unknown 180896 5.153924
White 1932645 55.063166

Data visualization

The data visualization serves as a foundation for a more in-depth investigation by providing us with a preliminary understanding of the relationship between the variables (age, gender, ethnicity, and race) and the number of homeless people and how that relationship has evolved over time.

Total Homelessness by Year and Age Group

# line plot for "Total Homelessness by Year and Age Group"
library(plotly)
## 
## Attaching package: 'plotly'
## The following object is masked from 'package:ggplot2':
## 
##     last_plot
## The following object is masked from 'package:stats':
## 
##     filter
## The following object is masked from 'package:graphics':
## 
##     layout
age_plot <- age_clean |>
  select(-location) |>
  group_by(calendar_year,age_group_public) |>
  summarise(total_homelessness = sum(experiencing_homelessness))
## `summarise()` has grouped output by 'calendar_year'. You can override using the
## `.groups` argument.
age_plot |> 
  mutate(text_label = str_c("Year: ", calendar_year, "\nAge Group: ", age_group_public)) |> 
  plot_ly(x = ~calendar_year, y = ~total_homelessness, type = "scatter", mode = "line", color = ~age_group_public, colors = "viridis", text = ~text_label, alpha = 0.8) |>
    layout(title = "Total Homelessness by Year and Age Group")

The above graph shows that the majority of homeless individuals who have received assistance are under the age of 18. The proportion of homeless people who have received assistance overall is higher among lower age groups. So we should pay attention to younger group to give them broader help. The overall trend of the homeless population receiving assistance is rising between 2021 and 2022, which is probably because of the potential effects of the COVID-19 pandemic.

Total Homelessness by Year and Ethnicity

# bar chart for "Total Homelessness by Year and ethnicity"
ethnicity_plot <- ethnicity_clean |>
  select(-location) |>
  group_by(calendar_year,ethnicity) |>
  summarise(total_homelessness = sum(experiencing_homelessness))
## `summarise()` has grouped output by 'calendar_year'. You can override using the
## `.groups` argument.
ethnicity_plot |> 
  mutate(text_label = str_c("Year: ", calendar_year, "\nEthnicity: ", ethnicity)) |>
  plot_ly(x = ~calendar_year, y = ~total_homelessness, type = "bar", color = ~ethnicity, colors = "viridis", text = ~text_label, alpha = 0.8) |>
  layout(
    title = "Total Homelessness by Year and ethnicity")

From the bar chart above we can see that people whose race are Hispanic/Latinx make up a large percentage of people receiving homeless response services, approximately 35%. However, based on the 2020 Census data, the Latino population in the United States totals 62.1 million, constituting 18.9% of the overall population. This demographic group stands as the second-largest ethnic or racial category in the country.So we need to focus on the vulnerable groups in ethnicity based on the problem of homelessness.

Total Homelessness by Year and Gender

# bar chart for "Total Homelessness by Year and gender"
gender_plot <- gender_clean |>
  select(-location) |>
  group_by(calendar_year,gender) |>
  summarise(total_homelessness = sum(experiencing_homelessness))
## `summarise()` has grouped output by 'calendar_year'. You can override using the
## `.groups` argument.
gender_plot  |> 
  mutate(text_label = str_c("Year: ", calendar_year, "\nGender: ", gender)) |>
  plot_ly(x = ~calendar_year, y = ~total_homelessness, type = "bar", color = ~gender, colors = "viridis", text = ~text_label, alpha = 0.8) |>
  layout(
    title = "Total Homelessness by Year and gender")

According to the bar chart by sex, we can know that the majority of people in need of assistance who are homeless are women, indicating that homelessness is a problem that affects more women than men and is rising annually for all genders from 2017 to 2022. The trend is downward and more controlled in 2023.

Total Homelessness by Year and Race

# line plot for "Total Homelessness by Year and race"
race_plot <- race_clean |>
  select(-location) |>
  group_by(calendar_year,race) |>
  summarise(total_homelessness = sum(experiencing_homelessness))
## `summarise()` has grouped output by 'calendar_year'. You can override using the
## `.groups` argument.
race_plot  |> 
  mutate(text_label = str_c("Year: ", calendar_year, "\nRace: ", race)) |>
  plot_ly(x = ~calendar_year, y = ~total_homelessness,  type = "scatter", mode = "line", color = ~race, colors = "viridis", text = ~text_label, alpha = 0.8) |>
  layout(
    title = "Total Homelessness by Year and gender" ,
    legend = list(
      x = 1,
      y = 0.5, 
      traceorder = "normal",
      bgcolor = "white", 
      bordercolor = "white",
      borderwidth = 0.5
    ))
# pie chart for "Percentage of Total Homeless by Race"
race_clean |>
  group_by(race) |>
  summarise(total_homelessness = sum(experiencing_homelessness)) |>
  mutate(percentage = total_homelessness / sum(total_homelessness)) |>
  plot_ly(labels = ~race, values = ~percentage, type = "pie", hole = 0.4) |>
   layout(title = "Percentage of Total Homeless by Race")

Based on the percantage of total homeless by race, we can see that white people accounted for the largest percentage of homeless people receiving help at 55.1%, followed by Black, Afircan American or African, the next largest group at 29.3%. This suggests that there is a relationship between race and Number of people receiving homelessness services, and that there is a need to focus on these two groups of people.

Kruskal-Wallis test

Next, we employ hypothesis testing as a method to more accurately assess the significance of relationships between variables. Through hypothesis testing, we aim to rigorously determine whether observed associations or differences between variables are statistically significant, providing a robust means to infer the presence of genuine relationships in our data.

When data does not meet the assumption of normality and involves multiple groups, we can consider using a non-parametric method for analysis, such as the Kruskal-Wallis test. The Kruskal-Wallis test is a non-parametric test used to compare differences among three or more groups, and it is less restrictive in terms of assumptions about the distribution of the data.

\(H_0\): the distribution of Y is the same across all groups.
\(H_1\): At least one group has a different distribution.

If the test result is significant, it indicates that at least one group’s median is different. However, it does not specify which groups are significantly different from each other. In such cases, you might need to perform multiple comparisons or follow-up pairwise tests to identify specific group differences.

# 'data' with 'Y' (continuous) and 'Group' (categorical)
kruskal.test(experiencing_homelessness ~ age_group_public, data = age_clean)
## 
##  Kruskal-Wallis rank sum test
## 
## data:  experiencing_homelessness by age_group_public
## Kruskal-Wallis chi-squared = 402.28, df = 7, p-value < 2.2e-16
kruskal.test(experiencing_homelessness ~ gender, data = gender_clean)
## 
##  Kruskal-Wallis rank sum test
## 
## data:  experiencing_homelessness by gender
## Kruskal-Wallis chi-squared = 469.88, df = 5, p-value < 2.2e-16
kruskal.test(experiencing_homelessness ~ ethnicity, data = ethnicity_clean)
## 
##  Kruskal-Wallis rank sum test
## 
## data:  experiencing_homelessness by ethnicity
## Kruskal-Wallis chi-squared = 311.31, df = 2, p-value < 2.2e-16
kruskal.test(experiencing_homelessness ~ race, data = race_clean)
## 
##  Kruskal-Wallis rank sum test
## 
## data:  experiencing_homelessness by race
## Kruskal-Wallis chi-squared = 630.47, df = 6, p-value < 2.2e-16
  • statistic: The test statistic (H).
  • parameter: The degrees of freedom (df) associated with the test.
  • p.value: The p-value of the test.

If the p-value is less than chosen significance level (0.05), we reject the null hypothesis.

According to the results, the p-value of experiencing_homelessness by gender is greater than 0.05, which means the distribution of number of experiencing homelessness is the same across gender groups. However, all other tests’ p-value is smaller than 0.01, which means at least one group has a different distribution in the whole age groups, ethnicity groups and race groups.

Part2: datasets “Hospital Encounters for Homeless Patients”

Description of datasets

While this dataset is rich in hospital encounter information, it seems to differ from the ‘demo’ datasets described earlier. The ‘demo’ datasets focus on broader statistics of people receiving homeless response services across various dimensions such as age, race, ethnicity, and gender, and are sourced from the Homelessness Data Integration System (HDIS) which compiles data from California’s Continuums of Care. In contrast, the provided dataset is more focused on hospital encounters of homeless individuals, offering a detailed view of healthcare utilization among this population. While this dataset is rich in hospital encounter information, it differs from the ‘demo’ datasets described earlier. In contrast, the provided dataset is more focused on hospital encounters of homeless individuals, offering a detailed view of healthcare utilization among this population.

  • Important Variables:

2019-2020-homeless-ip-and-ed-by-facility contains 17 variables, some of which are listed below:

Licensed Bed Size: the hospital’s number of licensed beds.
Homeless Indicator: indicates if the data is for Homeless or Non-Homeless encounters.
Demographic Category: Age, Race, Sex, or (Expected) Payer.
Encounters: Count of inpatient hospitalizations (i.e., discharges) or emergency department visits.
Total Hospital Encounters: Total inpatient hospitalizations or emergency department visits per hospital.

library(tidyverse)
library(knitr)
homeless_ip<-read.csv("2019-2020-homeless-ip-and-ed-by-facility.csv")
homeless_data<-homeless_ip %>% 
  filter(HomelessIndicator == "Homeless")

Summary

Again, To begin with the analysis process, we summarized the counts of Californian homeless people receiving appropriate health services from 2019 to 2020 of the 2019-2020-homeless-ip-and-ed-by-facility.csv dataset.

  • Important Variables:

EncounterType: whether receiving ED Visits or Inpatient Hospitalizations. FacilityName :name of hospitals and systems the hospitals belong to. Ownership: type of ownership that hospitals belong to.

Urban_Rural: whether hospitals locate in urban or rural areas.

LicensedBedSize: size of licensed beds per hospital.

PrimaryCareShortageArea: whether hospitals locate in area with limited primary care.

MentalHealthShortageArea: whether hospitals locate in area with limited mental health care.

homeless_data_for_summary =
  homeless_data |>
  filter(HomelessIndicator == "Homeless") |>
  select(EncounterType, FacilityName, System, Ownership, Urban_Rural, LicensedBedSize, PrimaryCareShortageArea, MentalHealthShortageArea, Encounters)

homeless_data_for_summary |>
  group_by(EncounterType) |>
  summarize(Homelessness = sum(Encounters)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
EncounterType Homelessness Percent
ED Visits 2633691 73.49042
Inpatient Hospitalizations 950029 26.50958 
homeless_data_for_summary |>
  group_by(FacilityName) |>
  summarize(Homelessness = sum(Encounters)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100)
## # A tibble: 425 × 3
##    FacilityName                         Homelessness  Percent
##    <chr>                                       <int>    <dbl>
##  1 AHMC Anaheim Regional Medical Center         6657 0.186   
##  2 AHMC Seton Medical Center                    4176 0.117   
##  3 AHMC Seton Medical Center Coastside            20 0.000558
##  4 Adventist Health Bakersfield                 8231 0.230   
##  5 Adventist Health Clearlake                   1572 0.0439  
##  6 Adventist Health Delano                       944 0.0263  
##  7 Adventist Health Glendale                    8072 0.225   
##  8 Adventist Health Hanford                     5212 0.145   
##  9 Adventist Health Howard Memorial              872 0.0243  
## 10 Adventist Health Lodi Memorial               8216 0.229   
## # ℹ 415 more rows
homeless_data_for_summary |>
  group_by(Ownership) |>
  summarize(Homelessness = sum(Encounters)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
Ownership Homelessness Percent
Government 839016 23.41187
Investor 730916 20.39545
Non-Profit 2013788 56.19267 
homeless_data_for_summary |>
  group_by(Urban_Rural) |>
  summarize(Homelessness = sum(Encounters)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
Urban_Rural Homelessness Percent
Rural/Frontier 170242 4.750427
Urban 3413478 95.249573 
homeless_data_for_summary |>
  group_by(LicensedBedSize) |>
  summarize(Homelessness = sum(Encounters)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
LicensedBedSize Homelessness Percent
1-99 208301 5.812424
100-199 744056 20.762113
200-299 782768 21.842331
300-399 814065 22.715642
400+ 1034530 28.867490 
homeless_data_for_summary |>
  group_by(PrimaryCareShortageArea) |>
  summarize(Homelessness = sum(Encounters)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
PrimaryCareShortageArea Homelessness Percent
No 2846216 79.42071
Yes 737504 20.57929 
homeless_data_for_summary |>
  group_by(MentalHealthShortageArea) |>
  summarize(Homelessness = sum(Encounters)) |>
  mutate(Percent = (Homelessness / sum(Homelessness)) * 100) |> knitr::kable()
MentalHealthShortageArea Homelessness Percent
No 2619228 73.08685
Yes 964492 26.91315

Data Visualization

Homeless Encounters by Demographic Group (Age, Payer, Race, Sex)

combined_plot <- plot_ly(homeless_data, x = ~DemographicValue, y = ~Encounters, 
                         type = 'bar', color = ~DemographicValue, 
                         colors = 'viridis') %>%
  add_bars(width = 1) %>%
  layout(title = "Homeless Encounters by Demographic Group",
         xaxis = list(title = "Demographic Value", tickangle = 45),
         yaxis = list(title = "Number of Homeless Encounters"),
         barmode = 'group')
combined_plot

This analysis provides a breakdown of homeless encounters by different demographic facets and indicates which groups have higher or lower counts of encounters. It suggests that middle-aged individuals, those covered by Medi-Cal, white individuals, and males have higher numbers of homeless encounters. These findings could point towards specific demographic groups that may require more focused services and interventions.

Homeless Encounters by Bed Size

plot2 <- plot_ly(homeless_data, x = ~LicensedBedSize, y = ~Encounters, 
                 type = 'bar', color = ~LicensedBedSize, 
                 colors = 'viridis') %>%
  layout(title = "Homeless Encounters by Bed Size",
         xaxis = list(title = "Licensed Bed Size", tickangle = 45),
         yaxis = list(title = "Number of Homeless Encounters"))
plot2

Among the variable Bed Size, the category of 400+ Licensed Bed Size shows a high number of encounters of homelessness as compared to other groups, indicating that larger hospitals may have a higher number of such encounters. This could be due to a larger capacity to serve more patients or a higher likelihood of being located in urban areas where homelessness rates may be higher. The second leading category is the 100-199 Licensed Bed Size which was relatively high, because hospitals with 100-199 beds might be specialized in services that are more frequently utilized by the homeless population, such as mental health or substance abuse treatment and these medium-sized hospitals could be strategically located in areas where the homeless population is higher.

Homeless Encounters by Area

plot3 <- plot_ly(homeless_data, x = ~Urban_Rural, y = ~Encounters, 
                 type = 'bar', color = ~Urban_Rural, 
                 colors = 'viridis') %>%
  layout(title = "Homeless Encounters in Urban vs. Rural Areas",
         xaxis = list(title = "Urban/Rural"),
         yaxis = list(title = "Number of Homeless Encounters"))

plot3

This plot help us gain the understanding of the relationship between the number of homelessness encounters and rural/ urban settlement. It is evident that in urban areas there are many people who tend to be homeless due to various reasons such as poverty. Urban hospitals are often larger and have more comprehensive services, which might also contribute to the higher number of encounters. On the other hand, in rural areas less people tend to be homeless since life there is cheap and easily affordable. The more the population in a certain place the higher the chances of homelessness encounters.

Homeless Encounters by Ownership Type

plot4 <- plot_ly(homeless_data, x = ~Ownership, y = ~Percent, 
                 type = 'bar', color = ~Ownership, 
                 colors = 'viridis') %>%
  layout(title = "Percentage of Homeless Encounters by Ownership Type",
         xaxis = list(title = "Ownership"),
         yaxis = list(title = "Percentage of Homeless Encounters"))
plot4

When comparing the percentage of homeless encounters by ownership the non-profit category showed the highest percentage in homeless encounters. Non-Profit hospitals may have more encounters with homeless individuals possibly due to their mission-driven approach, which may include providing care to underserved populations. Investor-owned hospitals, while having a higher percentage of encounters than government hospitals, may still be less than Non-Profit hospitals, potentially due to different operational goals and priorities. Government-owned hospitals having the lowest percentage could be a result of factors such as location, size, the scope of services offered, or specific governmental policies and funding for homeless services.